Cadmore media player playing video PS26: What's your strategy if there is no platform, REDUX
Video Player
Accessibility
Using forms mode, you can utilize keyboard shortcuts to make usability more efficient.
Space bar will play and pause the video. Pressing escape will close all displays and pressing it again will jump to the end of the player.
You can press A to open the Keyboard Shortcut Dialog which also provides an audio description of the player and all of its features.
Language
1x
Play Speed
Adjust Volume
Transcript
Search
Result Count
of
Click to jump to result
Search
Re-center
Segments
Result Count
of
Click to jump to result
Search
Name:
PS26: What's your strategy if there is no platform, REDUX
Description:
PS26: What's your strategy if there is no platform, REDUX
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/4b3ff2b5-4ac2-4f9a-ba35-971efb7c1a45/thumbnails/4b3ff2b5-4ac2-4f9a-ba35-971efb7c1a45.png
Duration:
T00H48M32S
Embed URL:
https://stream.cadmore.media/player/4b3ff2b5-4ac2-4f9a-ba35-971efb7c1a45
Content URL:
https://cadmoreoriginalmedia.blob.core.windows.net/4b3ff2b5-4ac2-4f9a-ba35-971efb7c1a45/PS26_REDUX.mp4?sv=2019-02-02&sr=c&sig=aXwzSMFPtbRKejJhefiNb1cuH5GcIi%2BIck8D62FPztg%3D&st=2026-10-08T22%3A30%3A06Z&se=2026-10-09T00%3A35%3A06Z&sp=r
Upload Date:
2026-09-30T00:00:00.0000000
Transcript:
Language: EN.
Segment:0 .
WILL: So as I mentioned this morning, we are revisiting a discussion we started two years ago on this stage. And that discussion is what is your strategy is if there is no platform. And I am joined by four colleagues here on the stage. And I'm going to ask to introduce themselves very briefly, but just as a reminder of the conversation we're intending to have. Two years, ago, we asked the uncomfortable question, what happens if platforms as we know it disappears.
WILL: Today those conversations have become very real. AI systems synthesize answers from across the published literature without sending users to our publishers platforms. Discovery is happening in different ways, and all of us have data that suggests our world and the role of infrastructure is changing rapidly. So with that, I'm going to ask each one of my panelists to introduce themselves, starting with Alison.
ALISON: Great. I'm Alison Mudditt. I'm the CEO of PLOS. Or I've been for nine years. Sorry, wrong button. And much longer than that in scholarly publishing in general. But we have been sort of grappling with a lot of the same problems, I think from a slightly different perspective of some of the others on this panel.
ALISON: So looking forward to the discussion.
WILL: Is that on?
ANDY: Is mine on? Yeah, mine works.
WILL: Yes.
ANDY: Hi, everyone. I'm Andy McGregor. I work with [? 2L. ?] I'm a VP of Product Management at Sage. And this problem is something that I suffer from daily, I think. It's pretty much the major thing on my desk at the moment is this particular strategy. So I'm looking forward to this discussion. Hi, I'm Ann Michael and I am currently independent. I, up until recently, was working at the American Institute of Publishing, physics publishing, as their chief transformation officer.
ANDY: Briefly worked for Alison too, which was fun. But I've been in and around scholarly publishing now for about 20 years, and have been constantly one of those folks that have tried to push forward and ask really annoying questions. So don't be unkind, Will. I won't. I'll try. Hi, I'm David Crotty.
ANDY: For the last year or so, I've been the executive director of Cold Spring Harbor Laboratory press. We're a small, not for profit life sciences publishing organization that's part of a research institution on Long Island. We publish about eight journals and a books program. Before that, I've done a little of everything. I was with Clark and Esposito for about four years as a consultant, Oxford University Press before that, Cold Spring Harbor Laboratory press before that, and before that, I was a research biologist doing developmental neuroscience research.
WILL: Thank you. So we've given our panelists a safe word. We're going to try to have a really candid, thoughtful discussion, maybe a little bit salty, a little bit edgy, but they reserve the right not to answer anything that I ask them. And we ask that you all be kind as well. Just one guardrail for our conversation. We aren't litigating. We're actually talking about changes to platforms and content delivery here.
WILL: So here's the first question. Andy, in the prep call for this panel, you mentioned Sage and your colleagues ran through, I think, about 13 different scenarios for the future of Sage's platforms. And when your senior leaders rank them by feasibility and desirability, status quo came out on top. Is that a failure of imagination or is it the best answer you have today?
ANDY: I think it's both. It was really galling. I spent a long time designing those 13 different scenarios from everything from the completely new journal models to New publication processes to new teaching models, and then to have an AI summit for senior leaders ranked status quo as number one. It was a little irritating, I have to honest. But I have since ran that same chat session with multiple people throughout Sage and status quo is always near the top when we start to rank the scenarios.
ANDY: And so I've had to get over my little strops that people are choosing the most boring scenario because I think it's true that status quo will remain important to us. People are still going to need to come to our sites to access the articles, to download the article. There may be less of them, there may be less of an important route, but we're still going to have to design for that route.
ANDY: It's still going to be a major issue for us. It's just now that there's at least 12 other scenarios we have to think design for as well. And it's not clear to me still which of those other 12 I need to put my efforts into.
WILL: Yes, how do you prioritize? So I'm curious, if the rest of you held votes with your management teams or colleagues, would you have had a similar outcome?
ALISON: So I think PLOS is actually coming from a different starting point, which is as an open access publisher whose content is all published CC BY, we've never been a platform centric publisher. So the changes that we see in the ecosystem are ones that I think are-- they are accelerated by AI, but the demand for us to change the way in which we serve our scientific communities was there long before AI.
DAVID: I'll say, as someone who works for a research institution, one of our strengths is computational biology. And there is huge enthusiasm for AI. It is going to revolutionize the way a lot of science is done using huge data sets, looking at things on a systems level instead of a cellular or molecular level. It's really going to change a huge amount of science. And that's how they're thinking about it and are enthusiastically adopting it.
DAVID: And it's almost a different set of problems and a different set of approaches than what we're dealing with. And what we're dealing with is how does it change the communication of science. And that's almost more of a piece with how is AI changing broader society, how do we gather information, how do we learn things, how do we teach people things more so than the practice of science.
DAVID: So I would say, they're saying, well, everything is completely different and revolutionized, but they're not thinking about necessarily us.
ANN: And first of all, I need to take a step back. I was remiss in saying that for basically 15 years, I ran a consulting company that I founded called Delta Think. And so I apologize to my colleagues here. It's funny. After all these years, you still get nervous when you're sitting up in front of people. But as part of that process of running that company and then working with other organizations, it's been really clear that this industry in general is very resilient and relies on certain things staying the same and has some insulation from things changing rapidly.
ANN: The subscription model, the fact that it takes three years to publish research, all these things that when a change hits, it takes a while to actually feel the impact of that change. So the status quo thing does not surprise me at all. And no offense, but sometimes I feel like we're an industry of frogs and water that's getting hotter and hotter. And I actually am very indebted to AI because I think some of the frogs are jumping out of the pot.
ANN: And I'm very happy to see that. [LAUGHTER]
WILL: Our intention wasn't to boil anyone today. [LAUGHTER] But one thing has changed for certain. And Ann, I'm coming back to you for this. Non-human readers now outnumber human readers on our no platforms. So as we think about that change, with the Silverchair Platform, we're building a new layer just for those non-human readers. They have different needs.
WILL: Should we stop building features for human readers? Are there different considerations we need to make of our technology, how we make our budget and resource investment decisions?
ANN: So I think this is a super interesting question. And I think one thing that we often do as a species is like we get something and we define one bucket without differentiation. And so I would say that even within non-human readers, there are different things. And some of those are orchestrated by humans who are on the other side of that process. And one thing we were talking about before thinking about, well, do anything for readers or not is, well, maybe.
ANN: And someone else said this about experimentation. I think David Sampson might have said something about this, too, that we don't have the option of sitting still. Maybe we need to structure some experiments thinking about, OK, let's suppose someone has consumed our content in an environment where it was summarized, and now they're actually heading over to our site. Is that person and that use case fundamentally different from when they got 25 links from Google or ScholarOne or something else, and do they need something different than what we've provided them before?
ANN: I want to go into inquiry mode. Well, I want to say, OK, so there's bots coming that are doing training. What do they need? There's application layer tools that are coming. What do they need? There are people on the other side of some of these things. What do they need?
ANN: And then who knows, maybe six or seven people are actually just putting in the URL, do we need to worry about them. So it's kind of not an answer, but I think the answer is I think it's too early to say we don't. And it's probably too early to say we always will.
WILL: Anybody else want to comment on how they're thinking about human users within their technology on their platforms?
DAVID: I mean, I was just thinking, I think I've reached a point where the journal homepage is largely designed and maintained to make the owners of the journal and the editors happy more so than the readers.
ANN: For other publishers to look at.
DAVID: Basically, yeah. I mean, I think there's a lot of window dressing that we still have, which is not to say we still don't have human readers and there are potential improvements we could make for them. But I think there's a lot that we do that we're still back in that print paradigm of you're going to have a cover and it's going to and a masthead and it's going to look like this. And I think the journal home page, for example, is one of those pieces.
ALISON: No? Oh, yeah. Just-- oh, sorry. No, I was just going to say, I think one of the elements that we're really starting to engage with now as we think about the broader range of outputs and PLOS's future being very much thinking about a distributed platform where we are hosting articles, the other outputs, data, code, whatever it is, sits in the appropriate repository.
ALISON: We're thinking about how we are moving metadata back and forth, how things like trust signals travel, what the contextual layer looks like. And so all of that, we absolutely have to be thinking about both humans and machines and different needs as we're thinking about those different contexts.
ANDY: Yeah, I agree with everything everyone said. I think for us, we still need to concentrate on humans, at least for now. For how long, we don't know. But there's plenty of things we put time and effort into maintaining for humans that they just don't use, so that the vast majority of people come directly to content pages. That's where we're going to have to concentrate the effort for humans.
ANDY: But we also have search functions, which no one uses. And no one's going to use any more, I don't think so. I think it's going to be as much taking features away as it is concentrating on humans. I think there's a lot of things we can lose in order to concentrate on where it really matters, which I think fundamentally for humans is that content page.
WILL: I tried taking away on site search once when I was a publisher. And then I found out all of my copy editing and production colleagues used it.
ANN: And the editors.
ANDY: And librarians.
WILL: So, David, in our prep call, we were talking a little bit about Google 0 and the possibility that 90% of your traffic may vanish. And that was never really meaningful traffic or valuable traffic to you anyway. And what was of value were the four or five people who actually read the paper, and then maybe the one or two people who actually cited it. And many publishers in this room are seeing traffic declines. Some even double digit traffic declines.
WILL: Counter hasn't caught up, as we talked about this morning. And increasingly, publishers are getting questions from their librarians, their subscribers drivers about the usage changes, about the declines in the value for the subscription. So what are you telling your librarian counterparts?
DAVID: It's a good question, actually, when we talked about it. And this is my example of why I generally don't use AI. If you read the Scholarly Kitchen today, I had to sit down and write a Scholarly Kitchen blog post to actually understand what I was trying to say to you to explain it to you. So writing is thinking, I think, at least for me still. So the idea is I'm dividing the world, like the ancient Greek poet did, into foxes and hedgehogs.
DAVID: And foxes know a little bit about a lot of things, and hedgehogs know one really important thing. And so if you think of that in terms of reading, most of the reading that happens on our platform is fox reading. People come in, they look at an abstract, hey, this looks interesting, I'll download the PDF, and I'll read it when I have time. And you have a stack of papers on your desk that you'll never get to-- a hard drive full of PDFs that you'll never read.
DAVID: But the hedgehog reading is what really matters is this is directly relevant to the work I do. And it may be a very small number of readers, but I'm going to go through every figure, every dot in every figure and understand exactly what's going on here. And I think that the AI tools can probably take care of a lot of the fox reading for us, but it can't do the hedgehog reading. It can't give you that deep understanding of the field.
DAVID: Every older researcher you talk to will give you a 20 minute rant of kids today, I have to teach them to read the original literature. You can't get an understanding of the field from reading a review article. So what happens to that understanding if you're reading a summary of a review article? So the platform to get back to the subject of this talk is that all of our sales, or at least our subscription sales, are based on counter metrics, what are the numbers that we do.
DAVID: And if and if a huge portion of our numbers are fox reading, those go away, how do we prove our value to the library that hedgehog reading is really, really important? But it's not a numbers game anymore. So I think it's a real question of if counter is no longer the right number, cost per download is meaningless, then how do we demonstrate that value to a librarian.
ANN: I just want to add something what you were talking about and when we were talking pre this meeting. I was starting to think about it as the sales funnel. If you think about it-- I'm sorry if that's a crass analogy, but the idea that you had all these leads, these things at the top, they were your fox readers. But now what you're getting are qualified leads, people with a purpose, people that have a reason to consume.
ANN: And somehow, I don't know the answer. But you're right, somehow, we have to communicate or measure how much more valuable that basically what we've just done for this librarian's constituency is save them a lot of time. We've just basically said don't have to necessarily go into every foxhole to find the thing that you need. And there has to be some value to that too.
ANN:
WILL: Anyone else have a perspective thought? All right. Alison, you have something of a controversial view that Open Access was always meant to include machines, or meant open to machines, at least for PLOS. And you disagree with Open Access publishers who want AI companies to pay in some way. Folks who kind of disagree with that perspective would say using content for training is extractive or outside the bounds of creative commons licenses.
WILL: What's your perspective here?
ALISON: So I think the problem for me is that the question assumes that the economics work by AI extracting something. And I do think there are extractive elements here. But the way in which we at PLOS, the way in which I think about what we do at PLOS is that we are offering a service to the research enterprise, which is about disseminating, sharing the content, adding impact. And so institutions, funders, librarians pay for that because we are providing a service, not because there's some assumption of downstream that we are recovering the cost from someone who's using that output, whether it's to read or whether it's to train.
ALISON: And if you think about the way in which open access works under a CC BY license, it is deliberately free for anybody to reuse. And so if I then say, I want to go and run after this new category of user and ask them to pay me something when I'm not asking anyone else, it's kind of counterintuitive. I mean, it runs counter to the argument that I've just made about the purpose of Open Access. And CC BY was intended to be for reuse.
ALISON: Now the problem that we have that we don't have a clear solution to is that CC BY and open access have worked on the basis of attribution. And of course, that's the key piece of CC BY and Open Access has been stripped out by LLMs at this point in time. I don't think that's necessarily inevitable. I think it's a design flaw. And it is something that can be changed going forward.
ALISON: But that's really where our authors see it as a problem. Our authors are fine with reuse. They publish with us because they want their content to be reused. The problem is the lack of attribution. And so that's really, I think, the challenge that we're facing.
WILL: Interesting.
DAVID: I mean, commercial reuse is at the heart of Open Access. That has always been at the core. Why do governments fund research? Well, to improve the lives of the people in that country. And that includes economic improvements. That means new companies start, that means jobs, that means tax money, that means new things. And the reason cc-by is required is so that companies can take the things that governments are funding and turn those into things that make money and drive the economy of the country.
DAVID: So I don't think there's a question there whatsoever. But I think the other piece that comes up constantly is this question of attribution versus citation. And most of the researchers are worried about citation, which is an informal practice. CC BY says, if you republish this work, you need to attribute it to the original author. If you use the work that's reuse, if you use the work, as an academic person, you should probably cite where it comes from.
DAVID: But there's no legal license requirement that says, if I read Rob's paper, that I have to cite him. I can be mean and not cite, or cite my own paper because I'd rather get the glory of that. So it's almost separate issue of the legal licensing copyright rules versus the traditions of the academy and how we give credit and build careers.
WILL: Thank you. Shifting gears just a little bit. Andy, you said Sage's customers are asking what you're doing about MCP endpoints. And when you ask them what they need, they don't know. Someone an endpoint publisher, someone aggregation, similar to what Wiley is doing, some are talking about walled gardens, which is kind of Elsevier's approach. This is a really consequential architectural decision and it seems like it is being made without requirements.
WILL: What are you doing about this?
ANDY: Speaking to a lot of different librarians at the moment, and we're not rushing to build anything. I think that's the key thing here is that it's not clear what people want. We spoke to some librarians who are really enthusiastic hobbyists, using MCP endpoints to build amazing things, some of them using MCP endpoints to build incredibly complex Boolean searches, which I thought was the most librarian use case I've ever seen. I say that as a librarian, I was very impressed.
ANDY: Some of them are thinking, how do we improve our discovery services, some are thinking longer term, some of them are just exploring what MCP means for them. But no one sat down and said, we need this and we need this now. There doesn't seem to be, from the people we've spoken to, a burning need to reuse their content and enable their researchers to enable students in any way. So we're waiting and watching at the moment. We don't think there's a clear demand for anything we have.
ANDY: We are experimenting with lots of different things. We're working with Wiley on their gateway, we're working with Elsevier, we're working on building our own internal MCP endpoints to move content around for licensing deals, et cetera. So we're really placing our bets on everything at the moment because the main thing we're worried about is, unless we really understand, unless we start to build, unless we start to experiment, when it becomes clear what people want, we won't be ready to move unless we're actually doing that experimentation, because this is complex technology.
ANDY: And really, you only get to understand it by using it in depth. Understanding where it's good, understanding where it's bad, that comes from use of it. So we are experimenting widely and we're waiting to see what happens. We're talking to many people as possible, but unless someone here has got the answer, which I'd love to know, it's not clear to me what people want or what the endpoint is going to be yet.
WILL: David or Alison, how are you thinking about your architectural strategy, your product and technology approach for your houses?
ALISON: So, I mean, it's interesting. And I was talking to somebody about this earlier. When I got to PLOS, we built all of our technology internally. We're in the middle of building a preprint server. And one of my early questions was, well, why are we doing that. We don't need to host preprints. And I think that is really the approach that we've taken to platform that the actual journal platform, as Andy was just saying, is becoming less important, as in it needs fewer features at this point.
ALISON: People aren't coming to it for different reasons. They're getting there through different gateways. And a lot of our vision in terms of thinking about moving beyond articles, thinking of a full knowledge stack of research outputs connected to a piece of research is really based on a much more distributed model. And so the questions are really the ones I raised earlier about what does the metadata look like, what's the overall ecosystem infrastructure, how do we make sure that a retraction can travel from one source to another, what's the contextual layer that humans and machines need to be able to understand that.
ALISON: So somewhat different perspective.
DAVID: Yeah. I mean, we're part of a research institution. And I don't know if anyone in this room has noticed, research funding has been going through some shaky times. [LAUGHTER]
WILL: You don't say.
DAVID: So it's very difficult to go to the institution and say, hey, throw bucket million our way so we can play around with this stuff. So we have to think about potential outside partners or playing a waiting game until the moment is right. And we talked about this in our pre meeting was one of the things that really made me think about is what do we have that's unique. Everybody's got a bunch of journal articles. We can go license those to OpenAI or Claude or whoever.
DAVID: Same as everybody. We have eight journals. We're going to be this tiny little drop in the bucket compared to Elsevier's thousands of journals. But what do you have that's unique? So if you're a medical publisher in a particular field, maybe you have the key clinical practice guidelines. If you're an engineering publisher, well, maybe you have the standards that you publish that nobody else has.
DAVID: And for us, we have 50 years of laboratory manuals. Nobody else has the corpus. We publish the first molecular biology laboratory manual. So we have this that no one else has. So the first thing I've decided is let's take that piece of it out of anything we license because this is special and unique. So our journal article is fine, whatever. And thinking about it, a company had approached us, said, hey, we're building a tool to help scientists plan their experiments.
DAVID: You have all these manuals. This will be great for our tool. We'll give you a few thousand dollars a year for that, which was sort of like, well, one, this is the heart of your tool. And if it works really well, then our whole methods publishing program is gone. So we said no because they wanted to give the tool away for free.
DAVID: So that didn't help a lot. But started thinking about other approaches, do we find a partner who could actually build a tool with us and understanding this is the heart of the tool. So we should be an owner of that thing with you, or thinking getting back to MCP servers. Do we build an MCP server based on this content, this unique content, and then maybe that's sold as part of the journal subscription?
DAVID: We have a method called spring arbor protocols. Maybe you get it as part of your subscription. Maybe you get it as subscription plus. And it's open to whatever. If you're subscribing to your institution, then it can tap into it and things like that. And that, in a lot of ways, is really attractive because we know how to sell subscriptions. And it's not a new world in some ways.
DAVID: And it also keeps us in control of the content and we can shut it down whenever we want if we need to.
WILL: It would be awesome if you actually end up doing that, and you can share some results in future years. And in our prep call, you said anyone claiming a three to five year strategy should be run away from. And the real work over the next, I don't know, 6 to 12 months is learning and adapting without burning down the house. Andy, earlier, you talked about needing to ride many horses because you can't afford to skip the one that wins. David, you've talked in some contexts about being a fast follower by necessity.
WILL: Alison, in a prep call, you talked about shifting towards leading indicators rather than respective reporting, suggesting some of those indicators will tell you your strategy needs to change. All of you have different perspective, different size organizations. Going one by one, what advice do you have for the publishers in the room about their near-term strategy?
ANN: OK, since you went to me first, I'll go first. And that does definitely sound like me. And I think David touched on this. Wendy did this great thing that I love, which was the balance between moving too fast or moving too slowly. And I think that plays into this. So if I were running a publisher right now, what I would be doing is something similar to what Andy was saying, coming up with key strategies and saying, OK, what are the possible directions we can go in.
ANN: I don't know which these are going to be, but what would move us from one of these scenarios to another. So what are key signals that we should be watching? Then what would be our reaction in these different scenarios? And given that I can understand that, is there a common capability or a common need that exists regardless of the outcome and do I have it right now? Because if I don't, that's the first thing I need to do.
ANN: Because in these 12 different scenarios, there may be some degree of commonality of capability that is helpful. So I mean, that's a very concrete tactical way to say what should I be working on. But the other thing is this mentality, this idea that the most important capability that anybody has at this point in time is curiosity. And I think, actually, someone brought that back-- David, I keep talking about you-- today, too.
ANN: And I completely agree. Ask questions, be curious. Smaller targeted experiments, not things that cost you $10 million or $1 million. But how do you just get started and learn. And I don't think you can afford to sit it out. But if you really think that you're going to have a plan that's going to stay stable for three years, you're not living in the world that I seem to be inhabiting right now.
ANN: So the thing is, what can you find that works no matter what that you need to do, and then how do you put together a mechanism to identify and watch signals that show you or help you to anticipate whether you're moving from one of those scenarios to another.
ANN: Yeah, I agree with Ann pretty much perfectly. That's exactly pretty much describes exactly how we're approaching it in Sage. I think the thing I'd add is the importance of that hands on experimentation. And not just isolating that for a few people in the organization, making as many people as possible able to try that organization, try that experimentation, because the technology is so complicated, it's so counterintuitive.
ANN: Understanding where it's good and where it's bad, you have to do that by using it. And because it's so capable, you can actually do really quite impactful experiments for very little money and very little effort. And in doing that, you're not only learning, you're making yourself a better consumer because you'll be constantly bombarded by people telling you that their AI tool will solve your problem.
ANN: And the only way you can know if that's true or not is if you're an informed consumer of that, use of that technology, because it makes you an informed consumer. So as well as having the strategy thing, that ground up hands on experimentation is going to be incredibly important. Without that, I worry that we won't be able to move quickly when it becomes clear the direction we need to go in.
ANN: We will not be ready to move. We'll have to spend ages building up our knowledge, building up our teams without that constant experimentation. It takes time, it takes effort. We're having to cancel other things to make time for it. But it's really important, I think.
DAVID: I mean, for an organization of our size, there may not be any huge first mover advantage that we would miss out on. There's a point if we have a lot of quality science that we publish. And if someone finds a great new way to present that, we can hop on that train. And I was at the new directions meeting last week, and I think I horrified a lot of people at my table by telling the old joke that science progresses one funeral at a time.
DAVID: The Academy is incredibly slow moving. Remember, we went online and that's going to change the entire career and funding. Oh, well, now it's all phones. It's going to change it. And it changed behaviors certainly, but it hasn't necessarily changed the way the Academy works. And you have to remember the people who are in power who make those decisions got there because the current system works really well for them.
DAVID: And they've trained their students. They've trained their postdocs to do that as well. So there is an inertia there. And I suspect-- and we are already seeing. You talk to any researcher, they're using these tools. There is no question whatsoever. But how that actually translates into structural changes in the career structure, in the funding structure of research. We don't yet know.
DAVID: And so the things we have now are what they use. And I don't want to throw those away until we see what else is needed, I guess.
ALISON: Yeah, I think in the way in which we've been approaching things, I mean, PLOS has always been at the front. That's really our mission. That's why we're here. So a different approach to some of these issues. The way in which we've thought about moving towards these leading indicators has really been scientific in some ways. But, we started out at three or four years ago with a series of extensive series of structured interviews with stakeholders across the community, researchers, administrators, librarians, and funders to understand what they still value about the publishing process, what are the key things that they want to see survive, and what is no longer important to them.
ALISON: That really informed this sort of 18 month R&D process that we've been through. But I think one of the key learnings from this is you have to know what's coming, but you also have to be willing to act on it. And, I will own up to something [AUDIO OUT] the problems that were emerging at least a decade ago, I think, with APCs is the primary funding model for Open Access wouldn't do anything about it soon enough.
ALISON: And so you can have leading indicators, but the capability that goes with that is being willing to act on them. And that's really important too. And I think takes into something we haven't really touched on here, which is you have to build an organization that's capable of doing that. And that has real shifts into the kind of culture that you want to build, not just working practices.
ALISON: I mean, we've been through a big shift at PLOS over recent years and moving from a sort of waterfall project based organization to a lowercase agile, not the formal ceremonies, but working in shorter sprints in a much more iterative way. That is a huge shift for many organizations. But I think it's essential to survival in the world that's coming.
DAVID: And just to reiterate on that, I mean, I think a lot of it falls on mission and what is mission. And PLOS's mission has long been we're going to improve and change the way science is communicated. I think we all think on those lines. But there's also, being an old school publisher, we are stewards of fields. We are here quality, those sorts of things, that different priorities and it makes much more sense if your priority is building that change in the world, then that's your primary drive.
WILL: Trying not to be a frog in the pot. We've referenced the kudos taming the crocodile report a lot today. And it captures a lot of the anxious thoughts in the room. There's one note that they have that I think is worth testing here, which is publisher's is credibility engines rather than content hosts. And this is a pivot towards curation and stewardship rather than distribution.
WILL: So Alison, starting with you, you've talked about policies, work to move in a direction away from the article as the core unit towards connected artifacts. You started this conversation today with a desire. Saying you don't have a desire necessarily to host any of it. That's a description of a publisher without a platform in the sense that we use the word. Is that where it goes or where our industry goes?
WILL: And if it is where we go, what is the last function still standing that we have to perform?
ALISON: So I think I would just make one small change to that, which is not entirely without a platform for now. I think we will continue to host our journal articles. I mean, they are hosted elsewhere. But that is our core platform. We still see articles as being a sort of critical part of the process. So we're not suggesting that articles go anywhere anytime soon. But yes, so I think curation, quality control, all of the human judgment that comes with that, there is more of that process that I think can and will be outsourced to machines, but without taking the human out of the loop.
ALISON: So I think we've had a number of conversations with people today about where they are using and thinking about using AI in the peer review process. I think it's entirely unrealistic to think that AI isn't already being used in the peer review process, let alone what it looked like in the future. But I think we've all been aligned on that core human judgment of can you trust what's coming, how do you understand and trust what's coming out of the AI review of the machine.
ALISON: So I think that is a core piece of it. The other piece is the contextual layer. And I think how you tie all of these different pieces together, how they connect, that's important for the machines that are reading it. So a machine needs to understand the relationship of the data set that's sitting over in repository X to the article that's sitting on our platform. So I think that absolutely is a role there, but it's really about that trust, credibility, and judgment that I think is the core of what we need to not only continue, but really double down on going forward.
WILL: Andy, David, or Ann, what survives?
DAVID: Well, there's an article Michael Clark wrote in 2010 about why journals hadn't been disrupted yet.
ANN: Really long one?
DAVID: Yes. But he posited the different functions of a journal. And in 2010, he'd already said dissemination is done. That doesn't matter anymore. Still, designation and filtration are the pieces that matter. And there was a really interesting, really provocative blog post someone put out this week talking about what's going to happen with AI and tying back to an earlier session today.
DAVID: Research integrity is not going to be a problem, because it's going to be just as easy to generate an actual truthful, real piece of research with AI than a fake one. Why would you make something fake and wrong when you can for the same amount of minimal effort. And so we're going to be flooded with this incredible deluge of true but largely meaningless articles, doesn't advance the field.
DAVID: It's some trivial sort of thing. And text is going to become super cheap, basically. And the problem, and this author to be very provocative, sort of said, well, maybe we had this weird 60 year period from the 1970s to now where we started actually requiring peer review in journals. And maybe that period is over, because in that period, an anonymous researcher from some place you've never heard of could get a fair shake, could get their paper peer reviewed, and get into a really good journal.
DAVID: And maybe we're moving back to the earlier times where you had to know the editor, or you're a famous person, or you're from a really big institution. And are we moving back to this almost reputation based economy and the negative-- I mean, he points out, we had centuries this worked just fine. It's incredible levels of inequity, and it'll probably shut out a huge amount.
DAVID: But to get noticed, maybe you have to be this. And I would counter that saying, well, we already have a mechanism to raise-- you're going to have this whole layer of sludge down at the bottom. And in a lot of ways, we already have that layer. I've seen researchers say there's two literatures. There's the stuff that's published in the journals that I trust and that I know and that I pay attention to and that I actually read, and then there's all this other stuff that people are putting out just to put another bean in the bean counters bucket.
DAVID: It's going to get so much worse. And we have a mechanism through the journals that we can elevate and get that designation. Hey, this is important, my experts and my expert peer reviewers said, you should look at this. And I keep thinking of the old faculty of 1,000 before it became F 1,000 and thinking, is that actually suddenly a really new important thing that we could be thinking about. Here's your Spotify playlist from the top developmental biologists telling you what you should read this week.
ANN: It's funny. I've been thinking about this a lot, that we keep talking about what we do and how we do it, and how publishers do it, and how it could change, and how this might change and this might come from here and this might go from there. But earlier this year, in the space of 11 weeks, all of the labs introduced some AI scientists, some kind of mechanism, benchmark workshop, workbench-- that's the word I'm looking-- for science.
ANN: And then you've got things coming along like periodic labs and what was it, cusp and some of these other things. Have we consider the fact that they might just jump right over us? I mean, what if science is done in these environments? And you don't have to worry about the provenance because it's all hooked up together right there from the second it's happening. Where do we fit in that environment?
ANN: What's our job then? And I think that we need to think those really scary things, even if we don't want to because I'm not saying that's going to happen. I don't have a crystal ball. We don't know. Nobody knows what's going to happen. But one of those scenarios is probably they don't need us at all anymore.
ANN: It's the only thing holding it all together that they want an impact factor and is the only reason that we have journals because we couldn't really rate the efficacy or the soundness of a researcher. And now we have the scale to do that where you don't need the journal to say, this is an important researcher with an important thought, that there's more there. And again, I'm not saying it's going to happen, but I think we really need to think about what does the nuclear option look like.
DAVID: There are funders who are talking about have to use an electronic notebook, which will then just feed your data directly into an AI. And if someone wants to know what's going on in Ann's lab, they'll ask the AI and they'll write a paper or tell them that you know the story of it. I mean, I think it was [? Ashutosh ?] who's been talking a lot about the separating out mechanical cognition from judgment based cognition and that the AI tools are really good at the mechanical cognition, are the figure legends.
ANN: But if the humans are working in these workbenches, the human cognition is still there.
DAVID: Right, but there's still a judgment level of when you say, well, the AI is going to tell me this is a really important--
ANN: Does the judgment level have to be a journal?
DAVID: No, but it has to be human, I think.
ALISON: Somehow.
WILL: We're all jumping out of the pot.
ALISON: No, I mean, I keep looking at Theo and Ellie in the front here, because last time I saw Sarah Teigen, she was telling me about one of her recent trip to China and just what's happening in the chemistry labs there, and not a lot of humans doing not a lot of things.
DAVID: It varies field to field. Computer science is being completely remade. Mathematics are being completely remade. Wet bench biology is going to take longer because you got to raise the mice.
ANN: Who is it that just did a wet lab? Somebody just opened one of the-- I just read that yesterday. I'll have to look it up.
WILL: No idea.
DAVID: But the other thing I think about also is a thing like an LLM is all about. It's a word prediction machine, what is the next most likely word. And in some ways, that compresses everything down into the average. And what's interesting to me is thinking about the margins, the outliers. I think of Kary Mullis, who invented polymerase chain reaction basically while tripping on acid and surfing.
DAVID: There's no rule that's going to give you PCR, which revolutionized molecular biology. So if you use these tools, what are you losing if you're putting bringing everything to the middle?
ANN: But that's what we're learning right now. Every single person in this room has probably produced something from AI. And unlike what we request our authors to do, has not necessarily completely digested what it is, what it means, and how it works. Well, shame on you. But the point is, if you're doing this, you learn after a while that you're actually producing things beyond your own level to absorb it.
ANN: You actually start to pull it back and you start to say, anything I produce from Claude, from GPT, from whatever I'm doing, I have to own it, It has to be me, I have to get it. And if I don't get it, it doesn't go anywhere. And now you bring humans into a process where they have this augmentation. They're still humans. They're still making the calls. They can still do LSD and surf.
ANN: [LAUGHTER]
WILL: Not advisable.
DAVID: But it's foxes and hedgehogs again. And obviously, we're going to get these incredible efficiency boosts at the cost of understanding.
ANN: But I think we're talking about two use cases, two different ones. I'm talking about the hedgehogs. They have better tools to be hedgehogs. They don't have to just be foxes.
WILL: Let's hold that thought for one second. We have maybe three minutes left. And I just want to see if anybody in the room has a question for any of our panelists.
ANN: He's my favorite curmudgeon.
DAVID: You kids.
WILL: Any questions?
DAVID: I mean, I can argue more with Ann. So I mean, I have a kid who just graduated high school, starting college. So I've been thinking a lot about learning and how AI is impacting. All of his classes now, everything is in class writing or in class participation. There are no more research papers assigned. The research paper is over as far as college education goes. But process matters.
DAVID: And that's the thing I have of if I get efficient and I skip process to get to the output, then I haven't learned anything. So I wrote in the Scholarly Kitchen today, we have enough crappy papers about Hemingway. We don't need that output. The world has enough of them. But we need more people who go through the process of writing a crappy paper about Hemingway to learn how to think and to learn how to analyze things and to come out of that.
DAVID: And the output is irrelevant. The process is what matters. And the output of a graduate school is not a PhD thesis. It's a scientist who can think like a scientist. And that's what we need to be careful of.
ANN: But that's an education. There's a lot of changes that have to happen in education. So people learn how to use the tools that they have well enough and still remember and still understand and still learn the fundamental concepts. But that's an education issue.
DAVID: But we're creating a lot of shortcuts that make it much--
ANN: I love my calculator. I still know how to do math.
DAVID: Yeah, but you had to learn math before-- you had to learn to add before you had your calculator.
WILL: Andy or Alison, any closing thoughts?
DAVID: You guys can go home. We're going to continue.
WILL: Anything you want to share to wrap up?
ANDY: Nothing springing to mind, unfortunately.
WILL: Fair enough.
ANDY: Still digesting the Ann and David argument, to be honest.
DAVID: We'll be at the bar later.
ALISON: I'm going to end by just agreeing with David on something. Here we go. No, I was having an argument or debate over drinks with some friends at the weekend about the latest AI advances in maths and solving the equations. And I was making the case that it actually matters that we understand how the equation was solved, not just that it was solved, but how because it's how that we learn from.
DAVID: The mistakes you make along the way open up new avenues for you. If you start at the endpoint, all those are gone.
WILL: This has been an incredibly fun panel. I think I've been guiding panels for a really, really long time. I've never, in one session, covered acid, surfing, hedgehogs, boiling frogs, technology decisions, and you name it. So please join me in thanking our panelists today. [APPLAUSE] And we are headed to a 20 minute networking break. So please come back and be ready to go at 3:20 when we'll be having our hello from the other side discussion.
WILL: So back at 3:20.
Segment:0 .
WILL: So as I mentioned this morning, we are revisiting a discussion we started two years ago on this stage. And that discussion is what is your strategy is if there is no platform. And I am joined by four colleagues here on the stage. And I'm going to ask to introduce themselves very briefly, but just as a reminder of the conversation we're intending to have. Two years, ago, we asked the uncomfortable question, what happens if platforms as we know it disappears.
WILL: Today those conversations have become very real. AI systems synthesize answers from across the published literature without sending users to our publishers platforms. Discovery is happening in different ways, and all of us have data that suggests our world and the role of infrastructure is changing rapidly. So with that, I'm going to ask each one of my panelists to introduce themselves, starting with Alison.
ALISON: Great. I'm Alison Mudditt. I'm the CEO of PLOS. Or I've been for nine years. Sorry, wrong button. And much longer than that in scholarly publishing in general. But we have been sort of grappling with a lot of the same problems, I think from a slightly different perspective of some of the others on this panel.
ALISON: So looking forward to the discussion.
WILL: Is that on?
ANDY: Is mine on? Yeah, mine works.
WILL: Yes.
ANDY: Hi, everyone. I'm Andy McGregor. I work with [? 2L. ?] I'm a VP of Product Management at Sage. And this problem is something that I suffer from daily, I think. It's pretty much the major thing on my desk at the moment is this particular strategy. So I'm looking forward to this discussion. Hi, I'm Ann Michael and I am currently independent. I, up until recently, was working at the American Institute of Publishing, physics publishing, as their chief transformation officer.
ANDY: Briefly worked for Alison too, which was fun. But I've been in and around scholarly publishing now for about 20 years, and have been constantly one of those folks that have tried to push forward and ask really annoying questions. So don't be unkind, Will. I won't. I'll try. Hi, I'm David Crotty.
ANDY: For the last year or so, I've been the executive director of Cold Spring Harbor Laboratory press. We're a small, not for profit life sciences publishing organization that's part of a research institution on Long Island. We publish about eight journals and a books program. Before that, I've done a little of everything. I was with Clark and Esposito for about four years as a consultant, Oxford University Press before that, Cold Spring Harbor Laboratory press before that, and before that, I was a research biologist doing developmental neuroscience research.
WILL: Thank you. So we've given our panelists a safe word. We're going to try to have a really candid, thoughtful discussion, maybe a little bit salty, a little bit edgy, but they reserve the right not to answer anything that I ask them. And we ask that you all be kind as well. Just one guardrail for our conversation. We aren't litigating. We're actually talking about changes to platforms and content delivery here.
WILL: So here's the first question. Andy, in the prep call for this panel, you mentioned Sage and your colleagues ran through, I think, about 13 different scenarios for the future of Sage's platforms. And when your senior leaders rank them by feasibility and desirability, status quo came out on top. Is that a failure of imagination or is it the best answer you have today?
ANDY: I think it's both. It was really galling. I spent a long time designing those 13 different scenarios from everything from the completely new journal models to New publication processes to new teaching models, and then to have an AI summit for senior leaders ranked status quo as number one. It was a little irritating, I have to honest. But I have since ran that same chat session with multiple people throughout Sage and status quo is always near the top when we start to rank the scenarios.
ANDY: And so I've had to get over my little strops that people are choosing the most boring scenario because I think it's true that status quo will remain important to us. People are still going to need to come to our sites to access the articles, to download the article. There may be less of them, there may be less of an important route, but we're still going to have to design for that route.
ANDY: It's still going to be a major issue for us. It's just now that there's at least 12 other scenarios we have to think design for as well. And it's not clear to me still which of those other 12 I need to put my efforts into.
WILL: Yes, how do you prioritize? So I'm curious, if the rest of you held votes with your management teams or colleagues, would you have had a similar outcome?
ALISON: So I think PLOS is actually coming from a different starting point, which is as an open access publisher whose content is all published CC BY, we've never been a platform centric publisher. So the changes that we see in the ecosystem are ones that I think are-- they are accelerated by AI, but the demand for us to change the way in which we serve our scientific communities was there long before AI.
DAVID: I'll say, as someone who works for a research institution, one of our strengths is computational biology. And there is huge enthusiasm for AI. It is going to revolutionize the way a lot of science is done using huge data sets, looking at things on a systems level instead of a cellular or molecular level. It's really going to change a huge amount of science. And that's how they're thinking about it and are enthusiastically adopting it.
DAVID: And it's almost a different set of problems and a different set of approaches than what we're dealing with. And what we're dealing with is how does it change the communication of science. And that's almost more of a piece with how is AI changing broader society, how do we gather information, how do we learn things, how do we teach people things more so than the practice of science.
DAVID: So I would say, they're saying, well, everything is completely different and revolutionized, but they're not thinking about necessarily us.
ANN: And first of all, I need to take a step back. I was remiss in saying that for basically 15 years, I ran a consulting company that I founded called Delta Think. And so I apologize to my colleagues here. It's funny. After all these years, you still get nervous when you're sitting up in front of people. But as part of that process of running that company and then working with other organizations, it's been really clear that this industry in general is very resilient and relies on certain things staying the same and has some insulation from things changing rapidly.
ANN: The subscription model, the fact that it takes three years to publish research, all these things that when a change hits, it takes a while to actually feel the impact of that change. So the status quo thing does not surprise me at all. And no offense, but sometimes I feel like we're an industry of frogs and water that's getting hotter and hotter. And I actually am very indebted to AI because I think some of the frogs are jumping out of the pot.
ANN: And I'm very happy to see that. [LAUGHTER]
WILL: Our intention wasn't to boil anyone today. [LAUGHTER] But one thing has changed for certain. And Ann, I'm coming back to you for this. Non-human readers now outnumber human readers on our no platforms. So as we think about that change, with the Silverchair Platform, we're building a new layer just for those non-human readers. They have different needs.
WILL: Should we stop building features for human readers? Are there different considerations we need to make of our technology, how we make our budget and resource investment decisions?
ANN: So I think this is a super interesting question. And I think one thing that we often do as a species is like we get something and we define one bucket without differentiation. And so I would say that even within non-human readers, there are different things. And some of those are orchestrated by humans who are on the other side of that process. And one thing we were talking about before thinking about, well, do anything for readers or not is, well, maybe.
ANN: And someone else said this about experimentation. I think David Sampson might have said something about this, too, that we don't have the option of sitting still. Maybe we need to structure some experiments thinking about, OK, let's suppose someone has consumed our content in an environment where it was summarized, and now they're actually heading over to our site. Is that person and that use case fundamentally different from when they got 25 links from Google or ScholarOne or something else, and do they need something different than what we've provided them before?
ANN: I want to go into inquiry mode. Well, I want to say, OK, so there's bots coming that are doing training. What do they need? There's application layer tools that are coming. What do they need? There are people on the other side of some of these things. What do they need?
ANN: And then who knows, maybe six or seven people are actually just putting in the URL, do we need to worry about them. So it's kind of not an answer, but I think the answer is I think it's too early to say we don't. And it's probably too early to say we always will.
WILL: Anybody else want to comment on how they're thinking about human users within their technology on their platforms?
DAVID: I mean, I was just thinking, I think I've reached a point where the journal homepage is largely designed and maintained to make the owners of the journal and the editors happy more so than the readers.
ANN: For other publishers to look at.
DAVID: Basically, yeah. I mean, I think there's a lot of window dressing that we still have, which is not to say we still don't have human readers and there are potential improvements we could make for them. But I think there's a lot that we do that we're still back in that print paradigm of you're going to have a cover and it's going to and a masthead and it's going to look like this. And I think the journal home page, for example, is one of those pieces.
ALISON: No? Oh, yeah. Just-- oh, sorry. No, I was just going to say, I think one of the elements that we're really starting to engage with now as we think about the broader range of outputs and PLOS's future being very much thinking about a distributed platform where we are hosting articles, the other outputs, data, code, whatever it is, sits in the appropriate repository.
ALISON: We're thinking about how we are moving metadata back and forth, how things like trust signals travel, what the contextual layer looks like. And so all of that, we absolutely have to be thinking about both humans and machines and different needs as we're thinking about those different contexts.
ANDY: Yeah, I agree with everything everyone said. I think for us, we still need to concentrate on humans, at least for now. For how long, we don't know. But there's plenty of things we put time and effort into maintaining for humans that they just don't use, so that the vast majority of people come directly to content pages. That's where we're going to have to concentrate the effort for humans.
ANDY: But we also have search functions, which no one uses. And no one's going to use any more, I don't think so. I think it's going to be as much taking features away as it is concentrating on humans. I think there's a lot of things we can lose in order to concentrate on where it really matters, which I think fundamentally for humans is that content page.
WILL: I tried taking away on site search once when I was a publisher. And then I found out all of my copy editing and production colleagues used it.
ANN: And the editors.
ANDY: And librarians.
WILL: So, David, in our prep call, we were talking a little bit about Google 0 and the possibility that 90% of your traffic may vanish. And that was never really meaningful traffic or valuable traffic to you anyway. And what was of value were the four or five people who actually read the paper, and then maybe the one or two people who actually cited it. And many publishers in this room are seeing traffic declines. Some even double digit traffic declines.
WILL: Counter hasn't caught up, as we talked about this morning. And increasingly, publishers are getting questions from their librarians, their subscribers drivers about the usage changes, about the declines in the value for the subscription. So what are you telling your librarian counterparts?
DAVID: It's a good question, actually, when we talked about it. And this is my example of why I generally don't use AI. If you read the Scholarly Kitchen today, I had to sit down and write a Scholarly Kitchen blog post to actually understand what I was trying to say to you to explain it to you. So writing is thinking, I think, at least for me still. So the idea is I'm dividing the world, like the ancient Greek poet did, into foxes and hedgehogs.
DAVID: And foxes know a little bit about a lot of things, and hedgehogs know one really important thing. And so if you think of that in terms of reading, most of the reading that happens on our platform is fox reading. People come in, they look at an abstract, hey, this looks interesting, I'll download the PDF, and I'll read it when I have time. And you have a stack of papers on your desk that you'll never get to-- a hard drive full of PDFs that you'll never read.
DAVID: But the hedgehog reading is what really matters is this is directly relevant to the work I do. And it may be a very small number of readers, but I'm going to go through every figure, every dot in every figure and understand exactly what's going on here. And I think that the AI tools can probably take care of a lot of the fox reading for us, but it can't do the hedgehog reading. It can't give you that deep understanding of the field.
DAVID: Every older researcher you talk to will give you a 20 minute rant of kids today, I have to teach them to read the original literature. You can't get an understanding of the field from reading a review article. So what happens to that understanding if you're reading a summary of a review article? So the platform to get back to the subject of this talk is that all of our sales, or at least our subscription sales, are based on counter metrics, what are the numbers that we do.
DAVID: And if and if a huge portion of our numbers are fox reading, those go away, how do we prove our value to the library that hedgehog reading is really, really important? But it's not a numbers game anymore. So I think it's a real question of if counter is no longer the right number, cost per download is meaningless, then how do we demonstrate that value to a librarian.
ANN: I just want to add something what you were talking about and when we were talking pre this meeting. I was starting to think about it as the sales funnel. If you think about it-- I'm sorry if that's a crass analogy, but the idea that you had all these leads, these things at the top, they were your fox readers. But now what you're getting are qualified leads, people with a purpose, people that have a reason to consume.
ANN: And somehow, I don't know the answer. But you're right, somehow, we have to communicate or measure how much more valuable that basically what we've just done for this librarian's constituency is save them a lot of time. We've just basically said don't have to necessarily go into every foxhole to find the thing that you need. And there has to be some value to that too.
ANN:
WILL: Anyone else have a perspective thought? All right. Alison, you have something of a controversial view that Open Access was always meant to include machines, or meant open to machines, at least for PLOS. And you disagree with Open Access publishers who want AI companies to pay in some way. Folks who kind of disagree with that perspective would say using content for training is extractive or outside the bounds of creative commons licenses.
WILL: What's your perspective here?
ALISON: So I think the problem for me is that the question assumes that the economics work by AI extracting something. And I do think there are extractive elements here. But the way in which we at PLOS, the way in which I think about what we do at PLOS is that we are offering a service to the research enterprise, which is about disseminating, sharing the content, adding impact. And so institutions, funders, librarians pay for that because we are providing a service, not because there's some assumption of downstream that we are recovering the cost from someone who's using that output, whether it's to read or whether it's to train.
ALISON: And if you think about the way in which open access works under a CC BY license, it is deliberately free for anybody to reuse. And so if I then say, I want to go and run after this new category of user and ask them to pay me something when I'm not asking anyone else, it's kind of counterintuitive. I mean, it runs counter to the argument that I've just made about the purpose of Open Access. And CC BY was intended to be for reuse.
ALISON: Now the problem that we have that we don't have a clear solution to is that CC BY and open access have worked on the basis of attribution. And of course, that's the key piece of CC BY and Open Access has been stripped out by LLMs at this point in time. I don't think that's necessarily inevitable. I think it's a design flaw. And it is something that can be changed going forward.
ALISON: But that's really where our authors see it as a problem. Our authors are fine with reuse. They publish with us because they want their content to be reused. The problem is the lack of attribution. And so that's really, I think, the challenge that we're facing.
WILL: Interesting.
DAVID: I mean, commercial reuse is at the heart of Open Access. That has always been at the core. Why do governments fund research? Well, to improve the lives of the people in that country. And that includes economic improvements. That means new companies start, that means jobs, that means tax money, that means new things. And the reason cc-by is required is so that companies can take the things that governments are funding and turn those into things that make money and drive the economy of the country.
DAVID: So I don't think there's a question there whatsoever. But I think the other piece that comes up constantly is this question of attribution versus citation. And most of the researchers are worried about citation, which is an informal practice. CC BY says, if you republish this work, you need to attribute it to the original author. If you use the work that's reuse, if you use the work, as an academic person, you should probably cite where it comes from.
DAVID: But there's no legal license requirement that says, if I read Rob's paper, that I have to cite him. I can be mean and not cite, or cite my own paper because I'd rather get the glory of that. So it's almost separate issue of the legal licensing copyright rules versus the traditions of the academy and how we give credit and build careers.
WILL: Thank you. Shifting gears just a little bit. Andy, you said Sage's customers are asking what you're doing about MCP endpoints. And when you ask them what they need, they don't know. Someone an endpoint publisher, someone aggregation, similar to what Wiley is doing, some are talking about walled gardens, which is kind of Elsevier's approach. This is a really consequential architectural decision and it seems like it is being made without requirements.
WILL: What are you doing about this?
ANDY: Speaking to a lot of different librarians at the moment, and we're not rushing to build anything. I think that's the key thing here is that it's not clear what people want. We spoke to some librarians who are really enthusiastic hobbyists, using MCP endpoints to build amazing things, some of them using MCP endpoints to build incredibly complex Boolean searches, which I thought was the most librarian use case I've ever seen. I say that as a librarian, I was very impressed.
ANDY: Some of them are thinking, how do we improve our discovery services, some are thinking longer term, some of them are just exploring what MCP means for them. But no one sat down and said, we need this and we need this now. There doesn't seem to be, from the people we've spoken to, a burning need to reuse their content and enable their researchers to enable students in any way. So we're waiting and watching at the moment. We don't think there's a clear demand for anything we have.
ANDY: We are experimenting with lots of different things. We're working with Wiley on their gateway, we're working with Elsevier, we're working on building our own internal MCP endpoints to move content around for licensing deals, et cetera. So we're really placing our bets on everything at the moment because the main thing we're worried about is, unless we really understand, unless we start to build, unless we start to experiment, when it becomes clear what people want, we won't be ready to move unless we're actually doing that experimentation, because this is complex technology.
ANDY: And really, you only get to understand it by using it in depth. Understanding where it's good, understanding where it's bad, that comes from use of it. So we are experimenting widely and we're waiting to see what happens. We're talking to many people as possible, but unless someone here has got the answer, which I'd love to know, it's not clear to me what people want or what the endpoint is going to be yet.
WILL: David or Alison, how are you thinking about your architectural strategy, your product and technology approach for your houses?
ALISON: So, I mean, it's interesting. And I was talking to somebody about this earlier. When I got to PLOS, we built all of our technology internally. We're in the middle of building a preprint server. And one of my early questions was, well, why are we doing that. We don't need to host preprints. And I think that is really the approach that we've taken to platform that the actual journal platform, as Andy was just saying, is becoming less important, as in it needs fewer features at this point.
ALISON: People aren't coming to it for different reasons. They're getting there through different gateways. And a lot of our vision in terms of thinking about moving beyond articles, thinking of a full knowledge stack of research outputs connected to a piece of research is really based on a much more distributed model. And so the questions are really the ones I raised earlier about what does the metadata look like, what's the overall ecosystem infrastructure, how do we make sure that a retraction can travel from one source to another, what's the contextual layer that humans and machines need to be able to understand that.
ALISON: So somewhat different perspective.
DAVID: Yeah. I mean, we're part of a research institution. And I don't know if anyone in this room has noticed, research funding has been going through some shaky times. [LAUGHTER]
WILL: You don't say.
DAVID: So it's very difficult to go to the institution and say, hey, throw bucket million our way so we can play around with this stuff. So we have to think about potential outside partners or playing a waiting game until the moment is right. And we talked about this in our pre meeting was one of the things that really made me think about is what do we have that's unique. Everybody's got a bunch of journal articles. We can go license those to OpenAI or Claude or whoever.
DAVID: Same as everybody. We have eight journals. We're going to be this tiny little drop in the bucket compared to Elsevier's thousands of journals. But what do you have that's unique? So if you're a medical publisher in a particular field, maybe you have the key clinical practice guidelines. If you're an engineering publisher, well, maybe you have the standards that you publish that nobody else has.
DAVID: And for us, we have 50 years of laboratory manuals. Nobody else has the corpus. We publish the first molecular biology laboratory manual. So we have this that no one else has. So the first thing I've decided is let's take that piece of it out of anything we license because this is special and unique. So our journal article is fine, whatever. And thinking about it, a company had approached us, said, hey, we're building a tool to help scientists plan their experiments.
DAVID: You have all these manuals. This will be great for our tool. We'll give you a few thousand dollars a year for that, which was sort of like, well, one, this is the heart of your tool. And if it works really well, then our whole methods publishing program is gone. So we said no because they wanted to give the tool away for free.
DAVID: So that didn't help a lot. But started thinking about other approaches, do we find a partner who could actually build a tool with us and understanding this is the heart of the tool. So we should be an owner of that thing with you, or thinking getting back to MCP servers. Do we build an MCP server based on this content, this unique content, and then maybe that's sold as part of the journal subscription?
DAVID: We have a method called spring arbor protocols. Maybe you get it as part of your subscription. Maybe you get it as subscription plus. And it's open to whatever. If you're subscribing to your institution, then it can tap into it and things like that. And that, in a lot of ways, is really attractive because we know how to sell subscriptions. And it's not a new world in some ways.
DAVID: And it also keeps us in control of the content and we can shut it down whenever we want if we need to.
WILL: It would be awesome if you actually end up doing that, and you can share some results in future years. And in our prep call, you said anyone claiming a three to five year strategy should be run away from. And the real work over the next, I don't know, 6 to 12 months is learning and adapting without burning down the house. Andy, earlier, you talked about needing to ride many horses because you can't afford to skip the one that wins. David, you've talked in some contexts about being a fast follower by necessity.
WILL: Alison, in a prep call, you talked about shifting towards leading indicators rather than respective reporting, suggesting some of those indicators will tell you your strategy needs to change. All of you have different perspective, different size organizations. Going one by one, what advice do you have for the publishers in the room about their near-term strategy?
ANN: OK, since you went to me first, I'll go first. And that does definitely sound like me. And I think David touched on this. Wendy did this great thing that I love, which was the balance between moving too fast or moving too slowly. And I think that plays into this. So if I were running a publisher right now, what I would be doing is something similar to what Andy was saying, coming up with key strategies and saying, OK, what are the possible directions we can go in.
ANN: I don't know which these are going to be, but what would move us from one of these scenarios to another. So what are key signals that we should be watching? Then what would be our reaction in these different scenarios? And given that I can understand that, is there a common capability or a common need that exists regardless of the outcome and do I have it right now? Because if I don't, that's the first thing I need to do.
ANN: Because in these 12 different scenarios, there may be some degree of commonality of capability that is helpful. So I mean, that's a very concrete tactical way to say what should I be working on. But the other thing is this mentality, this idea that the most important capability that anybody has at this point in time is curiosity. And I think, actually, someone brought that back-- David, I keep talking about you-- today, too.
ANN: And I completely agree. Ask questions, be curious. Smaller targeted experiments, not things that cost you $10 million or $1 million. But how do you just get started and learn. And I don't think you can afford to sit it out. But if you really think that you're going to have a plan that's going to stay stable for three years, you're not living in the world that I seem to be inhabiting right now.
ANN: So the thing is, what can you find that works no matter what that you need to do, and then how do you put together a mechanism to identify and watch signals that show you or help you to anticipate whether you're moving from one of those scenarios to another.
ANN: Yeah, I agree with Ann pretty much perfectly. That's exactly pretty much describes exactly how we're approaching it in Sage. I think the thing I'd add is the importance of that hands on experimentation. And not just isolating that for a few people in the organization, making as many people as possible able to try that organization, try that experimentation, because the technology is so complicated, it's so counterintuitive.
ANN: Understanding where it's good and where it's bad, you have to do that by using it. And because it's so capable, you can actually do really quite impactful experiments for very little money and very little effort. And in doing that, you're not only learning, you're making yourself a better consumer because you'll be constantly bombarded by people telling you that their AI tool will solve your problem.
ANN: And the only way you can know if that's true or not is if you're an informed consumer of that, use of that technology, because it makes you an informed consumer. So as well as having the strategy thing, that ground up hands on experimentation is going to be incredibly important. Without that, I worry that we won't be able to move quickly when it becomes clear the direction we need to go in.
ANN: We will not be ready to move. We'll have to spend ages building up our knowledge, building up our teams without that constant experimentation. It takes time, it takes effort. We're having to cancel other things to make time for it. But it's really important, I think.
DAVID: I mean, for an organization of our size, there may not be any huge first mover advantage that we would miss out on. There's a point if we have a lot of quality science that we publish. And if someone finds a great new way to present that, we can hop on that train. And I was at the new directions meeting last week, and I think I horrified a lot of people at my table by telling the old joke that science progresses one funeral at a time.
DAVID: The Academy is incredibly slow moving. Remember, we went online and that's going to change the entire career and funding. Oh, well, now it's all phones. It's going to change it. And it changed behaviors certainly, but it hasn't necessarily changed the way the Academy works. And you have to remember the people who are in power who make those decisions got there because the current system works really well for them.
DAVID: And they've trained their students. They've trained their postdocs to do that as well. So there is an inertia there. And I suspect-- and we are already seeing. You talk to any researcher, they're using these tools. There is no question whatsoever. But how that actually translates into structural changes in the career structure, in the funding structure of research. We don't yet know.
DAVID: And so the things we have now are what they use. And I don't want to throw those away until we see what else is needed, I guess.
ALISON: Yeah, I think in the way in which we've been approaching things, I mean, PLOS has always been at the front. That's really our mission. That's why we're here. So a different approach to some of these issues. The way in which we've thought about moving towards these leading indicators has really been scientific in some ways. But, we started out at three or four years ago with a series of extensive series of structured interviews with stakeholders across the community, researchers, administrators, librarians, and funders to understand what they still value about the publishing process, what are the key things that they want to see survive, and what is no longer important to them.
ALISON: That really informed this sort of 18 month R&D process that we've been through. But I think one of the key learnings from this is you have to know what's coming, but you also have to be willing to act on it. And, I will own up to something [AUDIO OUT] the problems that were emerging at least a decade ago, I think, with APCs is the primary funding model for Open Access wouldn't do anything about it soon enough.
ALISON: And so you can have leading indicators, but the capability that goes with that is being willing to act on them. And that's really important too. And I think takes into something we haven't really touched on here, which is you have to build an organization that's capable of doing that. And that has real shifts into the kind of culture that you want to build, not just working practices.
ALISON: I mean, we've been through a big shift at PLOS over recent years and moving from a sort of waterfall project based organization to a lowercase agile, not the formal ceremonies, but working in shorter sprints in a much more iterative way. That is a huge shift for many organizations. But I think it's essential to survival in the world that's coming.
DAVID: And just to reiterate on that, I mean, I think a lot of it falls on mission and what is mission. And PLOS's mission has long been we're going to improve and change the way science is communicated. I think we all think on those lines. But there's also, being an old school publisher, we are stewards of fields. We are here quality, those sorts of things, that different priorities and it makes much more sense if your priority is building that change in the world, then that's your primary drive.
WILL: Trying not to be a frog in the pot. We've referenced the kudos taming the crocodile report a lot today. And it captures a lot of the anxious thoughts in the room. There's one note that they have that I think is worth testing here, which is publisher's is credibility engines rather than content hosts. And this is a pivot towards curation and stewardship rather than distribution.
WILL: So Alison, starting with you, you've talked about policies, work to move in a direction away from the article as the core unit towards connected artifacts. You started this conversation today with a desire. Saying you don't have a desire necessarily to host any of it. That's a description of a publisher without a platform in the sense that we use the word. Is that where it goes or where our industry goes?
WILL: And if it is where we go, what is the last function still standing that we have to perform?
ALISON: So I think I would just make one small change to that, which is not entirely without a platform for now. I think we will continue to host our journal articles. I mean, they are hosted elsewhere. But that is our core platform. We still see articles as being a sort of critical part of the process. So we're not suggesting that articles go anywhere anytime soon. But yes, so I think curation, quality control, all of the human judgment that comes with that, there is more of that process that I think can and will be outsourced to machines, but without taking the human out of the loop.
ALISON: So I think we've had a number of conversations with people today about where they are using and thinking about using AI in the peer review process. I think it's entirely unrealistic to think that AI isn't already being used in the peer review process, let alone what it looked like in the future. But I think we've all been aligned on that core human judgment of can you trust what's coming, how do you understand and trust what's coming out of the AI review of the machine.
ALISON: So I think that is a core piece of it. The other piece is the contextual layer. And I think how you tie all of these different pieces together, how they connect, that's important for the machines that are reading it. So a machine needs to understand the relationship of the data set that's sitting over in repository X to the article that's sitting on our platform. So I think that absolutely is a role there, but it's really about that trust, credibility, and judgment that I think is the core of what we need to not only continue, but really double down on going forward.
WILL: Andy, David, or Ann, what survives?
DAVID: Well, there's an article Michael Clark wrote in 2010 about why journals hadn't been disrupted yet.
ANN: Really long one?
DAVID: Yes. But he posited the different functions of a journal. And in 2010, he'd already said dissemination is done. That doesn't matter anymore. Still, designation and filtration are the pieces that matter. And there was a really interesting, really provocative blog post someone put out this week talking about what's going to happen with AI and tying back to an earlier session today.
DAVID: Research integrity is not going to be a problem, because it's going to be just as easy to generate an actual truthful, real piece of research with AI than a fake one. Why would you make something fake and wrong when you can for the same amount of minimal effort. And so we're going to be flooded with this incredible deluge of true but largely meaningless articles, doesn't advance the field.
DAVID: It's some trivial sort of thing. And text is going to become super cheap, basically. And the problem, and this author to be very provocative, sort of said, well, maybe we had this weird 60 year period from the 1970s to now where we started actually requiring peer review in journals. And maybe that period is over, because in that period, an anonymous researcher from some place you've never heard of could get a fair shake, could get their paper peer reviewed, and get into a really good journal.
DAVID: And maybe we're moving back to the earlier times where you had to know the editor, or you're a famous person, or you're from a really big institution. And are we moving back to this almost reputation based economy and the negative-- I mean, he points out, we had centuries this worked just fine. It's incredible levels of inequity, and it'll probably shut out a huge amount.
DAVID: But to get noticed, maybe you have to be this. And I would counter that saying, well, we already have a mechanism to raise-- you're going to have this whole layer of sludge down at the bottom. And in a lot of ways, we already have that layer. I've seen researchers say there's two literatures. There's the stuff that's published in the journals that I trust and that I know and that I pay attention to and that I actually read, and then there's all this other stuff that people are putting out just to put another bean in the bean counters bucket.
DAVID: It's going to get so much worse. And we have a mechanism through the journals that we can elevate and get that designation. Hey, this is important, my experts and my expert peer reviewers said, you should look at this. And I keep thinking of the old faculty of 1,000 before it became F 1,000 and thinking, is that actually suddenly a really new important thing that we could be thinking about. Here's your Spotify playlist from the top developmental biologists telling you what you should read this week.
ANN: It's funny. I've been thinking about this a lot, that we keep talking about what we do and how we do it, and how publishers do it, and how it could change, and how this might change and this might come from here and this might go from there. But earlier this year, in the space of 11 weeks, all of the labs introduced some AI scientists, some kind of mechanism, benchmark workshop, workbench-- that's the word I'm looking-- for science.
ANN: And then you've got things coming along like periodic labs and what was it, cusp and some of these other things. Have we consider the fact that they might just jump right over us? I mean, what if science is done in these environments? And you don't have to worry about the provenance because it's all hooked up together right there from the second it's happening. Where do we fit in that environment?
ANN: What's our job then? And I think that we need to think those really scary things, even if we don't want to because I'm not saying that's going to happen. I don't have a crystal ball. We don't know. Nobody knows what's going to happen. But one of those scenarios is probably they don't need us at all anymore.
ANN: It's the only thing holding it all together that they want an impact factor and is the only reason that we have journals because we couldn't really rate the efficacy or the soundness of a researcher. And now we have the scale to do that where you don't need the journal to say, this is an important researcher with an important thought, that there's more there. And again, I'm not saying it's going to happen, but I think we really need to think about what does the nuclear option look like.
DAVID: There are funders who are talking about have to use an electronic notebook, which will then just feed your data directly into an AI. And if someone wants to know what's going on in Ann's lab, they'll ask the AI and they'll write a paper or tell them that you know the story of it. I mean, I think it was [? Ashutosh ?] who's been talking a lot about the separating out mechanical cognition from judgment based cognition and that the AI tools are really good at the mechanical cognition, are the figure legends.
ANN: But if the humans are working in these workbenches, the human cognition is still there.
DAVID: Right, but there's still a judgment level of when you say, well, the AI is going to tell me this is a really important--
ANN: Does the judgment level have to be a journal?
DAVID: No, but it has to be human, I think.
ALISON: Somehow.
WILL: We're all jumping out of the pot.
ALISON: No, I mean, I keep looking at Theo and Ellie in the front here, because last time I saw Sarah Teigen, she was telling me about one of her recent trip to China and just what's happening in the chemistry labs there, and not a lot of humans doing not a lot of things.
DAVID: It varies field to field. Computer science is being completely remade. Mathematics are being completely remade. Wet bench biology is going to take longer because you got to raise the mice.
ANN: Who is it that just did a wet lab? Somebody just opened one of the-- I just read that yesterday. I'll have to look it up.
WILL: No idea.
DAVID: But the other thing I think about also is a thing like an LLM is all about. It's a word prediction machine, what is the next most likely word. And in some ways, that compresses everything down into the average. And what's interesting to me is thinking about the margins, the outliers. I think of Kary Mullis, who invented polymerase chain reaction basically while tripping on acid and surfing.
DAVID: There's no rule that's going to give you PCR, which revolutionized molecular biology. So if you use these tools, what are you losing if you're putting bringing everything to the middle?
ANN: But that's what we're learning right now. Every single person in this room has probably produced something from AI. And unlike what we request our authors to do, has not necessarily completely digested what it is, what it means, and how it works. Well, shame on you. But the point is, if you're doing this, you learn after a while that you're actually producing things beyond your own level to absorb it.
ANN: You actually start to pull it back and you start to say, anything I produce from Claude, from GPT, from whatever I'm doing, I have to own it, It has to be me, I have to get it. And if I don't get it, it doesn't go anywhere. And now you bring humans into a process where they have this augmentation. They're still humans. They're still making the calls. They can still do LSD and surf.
ANN: [LAUGHTER]
WILL: Not advisable.
DAVID: But it's foxes and hedgehogs again. And obviously, we're going to get these incredible efficiency boosts at the cost of understanding.
ANN: But I think we're talking about two use cases, two different ones. I'm talking about the hedgehogs. They have better tools to be hedgehogs. They don't have to just be foxes.
WILL: Let's hold that thought for one second. We have maybe three minutes left. And I just want to see if anybody in the room has a question for any of our panelists.
ANN: He's my favorite curmudgeon.
DAVID: You kids.
WILL: Any questions?
DAVID: I mean, I can argue more with Ann. So I mean, I have a kid who just graduated high school, starting college. So I've been thinking a lot about learning and how AI is impacting. All of his classes now, everything is in class writing or in class participation. There are no more research papers assigned. The research paper is over as far as college education goes. But process matters.
DAVID: And that's the thing I have of if I get efficient and I skip process to get to the output, then I haven't learned anything. So I wrote in the Scholarly Kitchen today, we have enough crappy papers about Hemingway. We don't need that output. The world has enough of them. But we need more people who go through the process of writing a crappy paper about Hemingway to learn how to think and to learn how to analyze things and to come out of that.
DAVID: And the output is irrelevant. The process is what matters. And the output of a graduate school is not a PhD thesis. It's a scientist who can think like a scientist. And that's what we need to be careful of.
ANN: But that's an education. There's a lot of changes that have to happen in education. So people learn how to use the tools that they have well enough and still remember and still understand and still learn the fundamental concepts. But that's an education issue.
DAVID: But we're creating a lot of shortcuts that make it much--
ANN: I love my calculator. I still know how to do math.
DAVID: Yeah, but you had to learn math before-- you had to learn to add before you had your calculator.
WILL: Andy or Alison, any closing thoughts?
DAVID: You guys can go home. We're going to continue.
WILL: Anything you want to share to wrap up?
ANDY: Nothing springing to mind, unfortunately.
WILL: Fair enough.
ANDY: Still digesting the Ann and David argument, to be honest.
DAVID: We'll be at the bar later.
ALISON: I'm going to end by just agreeing with David on something. Here we go. No, I was having an argument or debate over drinks with some friends at the weekend about the latest AI advances in maths and solving the equations. And I was making the case that it actually matters that we understand how the equation was solved, not just that it was solved, but how because it's how that we learn from.
DAVID: The mistakes you make along the way open up new avenues for you. If you start at the endpoint, all those are gone.
WILL: This has been an incredibly fun panel. I think I've been guiding panels for a really, really long time. I've never, in one session, covered acid, surfing, hedgehogs, boiling frogs, technology decisions, and you name it. So please join me in thanking our panelists today. [APPLAUSE] And we are headed to a 20 minute networking break. So please come back and be ready to go at 3:20 when we'll be having our hello from the other side discussion.
WILL: So back at 3:20.
Form Title Download
Transcript
Form Title Bookmark
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Form Title Share
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Form Title Accessibility and Keyboard Shortcuts
The cadmore media player is a professional video and audio player experience designed to deliver media as well as relevant transcript, segmentation, and metadata. On the screen is a traditional html video player with buttons such as play, pause, forward, captioning, language and play speed selection and more. Adjacent to the video player are generally tabs, one of which shows a scrolling transcript of the video and the other shows any segmentation of the media, sometimes referred to as chapters. At the top of the video is a menu bar containing an Explore button and Tools button. The Explore button allows you to dig deeper into the video by opening up an explore dialog which contains tabs for a visual navigation of the video using thumbnails, an about tab which contains rich metadata about the media, a segments tab which contains further information about the video, and a related tab which contains links to other media related to this media. The tools button opens a menu with options for sharing the media on social media, creating bookmarking links, creating embed codes for your website, generating citations, and more.
Keyboard Shortcuts
Keyboard shortcuts only work while you are in forms mode. By entering forms mode, you will be able to quickly access portions of the media player with single key commands.
Spacebar: Play/Pause Toggle
I: Info Menu,
O: Tools Menu
T: Transcript Tab,
S: Segments Tab
F: Forward 15,
R: Rewind 15
L: Languages,
P: Playspeed
M: Mute Toggle,
V: Volume Bar
N: Video Only Toggle,
B: Full Screen Toggle
C: Captions Toggle
Escape: Closes dialogs and menus / Exits Player
Z: Next Transcript Chapter
X: Previous Transcript Chapter
Q: Next Transcript Paragraph
W: Previous Transcript Paragraph
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Embed Size
-
You Size - Responsive
-
We Size - Responsive
-
Picture Overlay Popup
-
400 x 225
-
512 x 288
-
400 x 700
-
768 x 1024
-
1024 x 512
-
1280 x 608
Form Title Cite
No citation provided.
Info
Segments
Related
Thumbnails
Exit Clip Mode
You have requested to see a portion of the video outside of the clip.
Click OK to switch to the full video or click if you would like to stay inside the clip. You can always go back to the clip with Tools and Clip Mode